In the edge-defined, film-fed, crystal growth technique (the EFG process), tubular crystalline bodies, e.g., hollow bodies with round or polygonal cross-sections, are grown on a seed from a liquid film of feed material which is transported by capillary action from a crucible of molten silicon through one or more capillaries in a die to the end or top surface of the die. The shape of the crystalline body is determined by the external or edge configuration of the uppermost or top end surface of the die. Polygonally-shaped hollow bodies, e.g., "nonagons" or "octagons", grown by EFG are subdivided at their corners into a plurality of flat substrates that are used to form photovoltaic solar cells.
Known EFG dies of the type described and illustrated in U.S. Pat. No. 4,230,674 to Taylor et al. typically comprise an upper end surface, at least one capillary which intersects said upper end surface, and inner and outer side surfaces which intersect the upper end surface at about a 65.degree. angle, i.e., at about a 25.degree. angle to the vertical axis of the die.
With known EFG dies it tends to be difficult to initiate crystal growth, i.e. seed the crystal growth apparatus, for at least two reasons. First, crystal growth must be initiated within a relatively narrow range of die temperatures, i.e. the "growth window". Second, significant variations in temperature tend to exist around the circumference of the die.
The first of the above-discussed reasons contributes to some extent to the second of the above-discussed reasons. Crystal growth will not occur at a temperature below the growth window, i.e. the seed will freeze to the die. When crystal growth is attempted at a temperature above the narrow range of die temperatures where crystal growth is possible, the liquid film between the die and the seed (or growing crystal) will tend to rupture, i.e., the seed will break away from the liquid film, with the result that often the liquid silicon splashes onto or overflows the inner and/or outer side surfaces of the die. The spilled silicon tends to react to form silicon carbide deposits on the side surfaces of the die. Additionally, some silicon carbide will form on the side surfaces of the die in connection with the normal crystal growth process. These silicon carbide deposits, together with any unreacted liquid silicon on the side surfaces of the die, tend to change the thermal conductivity and emissivity of the die. Such changes in conductivity and emissivity will cause a localized change in the rate of loss of heat, with the result that the temperature around the circumference of the die is not even.
Thermal symmetry around the circumference of the die is further adversely affected by unavoidable subtle variations in the porosity, density and/or electrical conductivity of the graphite from which the die is made. These variations in graphite properties produce local variations in heat flow and cause variations in the depth to which spilled silicon will penetrate into the surfaces of the die, which in turn affects the temperature of the die.
Variations in thermal symmetry around the circumference of the die cause local changes in thickness of the growing crystalline body. Such variations in thickness tend to reduce the number of solar cells which can be produced from the hollow polygonal crystalline body, inasmuch as the thinner substrates cut from the crystalline body tend to be prone to breakage. Furthermore, the surfaces of substrates cut from crystalline bodies tend to be uneven due to the variations in thickness. This unevenness makes it more difficult to perform certain solar cell processing operations, such as attaching electrodes to the surfaces of the substrate.
As noted earlier, with known EFG crystal growth apparatus, when the menisci are ruptured, i.e., when the solid/liquid interface is terminated, liquid silicon tends to overflow the die, and such overflow may cause some flooding of some portions of the crystal growth apparatus adjacent the die. Such flooding of molten silicon results in thermal non-uniformities, as described above, and when excessive, can destroy the growth setup by fusing together the die and adjacent mechanical parts of the growth apparatus, and can possibly contaminate the molten silicon in the crucible and otherwise damage the crystal growth apparatus.
To avoid breakage of or interruption of the growth meniscus, it is essential for the human operator to use a high level of care and diligence in operating the growth apparatus. This high level of care is difficult for operators to maintain at all times, so that mishaps inevitably occur, with the result that the average useful life of EFG dies is significantly shortened by the mechanisms of non-uniform wetting and catastrophic flooding.
With known EFG dies there exists during crystal growth a large difference in temperature, or temperature gradient, between the growth meniscus at the top of the die and the bulk melt in the crucible, with the bulk melt having a much higher temperature than the melt in the growth meniscus. A large temperature gradient is desired in the growth meniscus itself to maintain stability in the crystal growth process. In known EFG dies a large temperature gradient in the growth meniscus results in a similar large temperature gradient in the die capillary between the growth meniscus and the melt in the crucible, due to continuity of heat flow through the die into the meniscus, and due to the high thermal resistance of the die.
An undesirable result of the required high temperature in the crucible can be the excess dissolution of the crucible material into the melt. As the melt flows up the die capillary from the crucible to the die top, the melt decreases in temperature and becomes supersaturated, so that the dissolved and reacted crucible materials will precipitate in the die capillaries and at the die tip. These precipitates can cause the capillary passage(s) to become clogged, preventing melt from reaching the die top, and thereby preventing further crystal growth. The precipitates can also alter the shape of the die top and the edges of the die top, thus changing the shape of the crystal grown from the die in undesirable ways.